The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input text for translation, creating a potential surface for indirect prompt injection where malicious instructions embedded in the source text could be passed to the output and influence downstream agents.
Ingestion points: The skill accepts user-supplied source text for translation into Thai or English as defined in SKILL.md.
Boundary markers: The instructions do not include specific boundary delimiters or warnings for the agent to ignore instructions embedded within the text being translated.
Capability inventory: The skill is capable of generating structured output in Markdown (.md) and Subtitle (.srt) formats. It does not utilize network or system-level tools.
Sanitization: The agent is explicitly instructed in SKILL.md to preserve markdown, HTML, and variables (e.g., {{name}}), which prevents the sanitization of potentially malicious structural content in the translation output.

translator-pro